Know where the heat is. Find the nearest shade and water.
Chhaon joins satellite images, weather and population into one heat map of Delhi. Then it shows each person the closest cooling point and a walking route.
- heat cells, 250 m each
- 42,387
- cooling points mapped
- 8,770
- open data sources
- 5
- walk used to measure access
- 10 min
Three steps from heat to action
Most people start with the map. Then they look for a place close to them.
- 01
See where the heat is
Open the heat map of Delhi. Red cells trap the most heat. Pick any day, also a past hot day.
Open the heat map - 02
Find the nearest cool place
Share your location. Get the 3 closest places with shade or water, and a walking route.
Find a place near me - 03
Plan which places to open
Officers get a ranked list of 10 places. The list shows how many more people it reaches.
See the daily plan
From satellite pixels to a walking route
Every part below runs today. Each part has one job.
Satellite surface heat
Landsat 8 and 9 thermal band, about 100 m. Read through Google Earth Engine.
Trees, buildings and water
Sentinel-2 images from AWS Open Data. We compute NDVI, NDBI and NDWI from them.
Weather of the day
Open-Meteo apparent temperature. It sets how severe the day is. The satellite sets where.
People per cell
WorldPop population is added to each cell. We rank by people at risk, not by area.
8,770 cooling points
Water points, parks, metro stations, hospitals, schools and places of worship from OpenStreetMap.
Walking routes
Amazon Location Service gives path and time. A failed route falls back to distance and direction.
Daily plan
A greedy pick of 10 places within 800 m. Coverage before and after is shown.
Replay any hot day
One date parameter replays a past heat day. Demo the plan in any season.
Community reports
Users report a closed place or wrong hours. We store each report for review.
Open, typed API
OpenAPI docs and a typed tRPC client. Other apps can reuse the data.
One score for each 250 m cell
The satellite shows where heat stays. The weather shows how bad the day is.
heat score = 60% satellite baseline + 40% weather of the day
- High heat: 70 and above
- Medium heat: 40 to 69
- Low heat: below 40
All values are surface heat from the last clear satellite pass. They are not air temperature.
Six AWS services in the live build
The API runs on Amazon EC2 in ap-southeast-2. Landsat surface temperature comes from Google Earth Engine.
Amazon EC2
Runs the API container.
Amazon ECR
Stores the container image.
Amazon RDS for PostgreSQL
Holds cells, points, weather, plans and reports.
Amazon Location Service
Calculates the walking route and time.
AWS IAM
Instance role. No keys in the image.
AWS Open Data
Hosts the Sentinel-2 images we read.
Not built yet. Ready to build.
We had two days. These items are planned. The design already has a place for each one.
- Next
All of Delhi NCR and other cities
The study area is one bounding box in the config. Change it and rerun the baseline job.
- Next
Automatic satellite refresh
A scheduled job reads each new Landsat pass. Raw scenes go to Amazon S3.
- Next
Official cooling zones
Import the lists of city authorities. Mark these places as official.
- Next
HTTPS and officer sign-in
Amazon CloudFront for HTTPS. Amazon Cognito for ward officers.
- Later
Heat alerts
Send SMS or WhatsApp alerts to people in high-heat cells. Use Amazon SNS.
- Later
Plain-language explainer
Amazon Bedrock explains in simple words why each place was picked.
- Later
Vulnerability layers
Add age, income and outdoor workers to the priority score. Today we use population only.
- Later
More Indian languages
The public app already speaks English and Hindi. Bengali, Punjabi and Urdu come next from the same copy file.